Characteristics of the movies in the last 30 years¶
In this report, we will analyze the data of movies from the last 30 years for Steven Spielberg for new movies. the purpose of the report is to consider the characteristics of popular movies. People's habit of movies may change along with time. So, we will organize the data first and filter out the invaluable data at first.¶
This time, the data is huge. we will analyze the relationship of the score and each element, such as vote, gross, and budget. we will understand these elements are contributing to the score or not. It is will important for us, for example, Does the high budget can get a high score? It may affect our analysis.¶
Then after we find the factor of the score, we can try to find in the last 30 years, which writers have more popular stories of movies. Then, finding a good performance of companies as well. the important thing we consider is popular star may contribute to the movies. So, we also analyze which star is popular in the last 30 years. In addition, we will focus on a specific period of time to analyze 2010~2017 that may get the data accurately.¶
We also will help Mr. Spielberg to review his recording of movies. We may can get some idea from the data in the past. We are very excited to work with Mr. Spielberg. He is the best director in the world. So, we will do our best to find what the data we need and give a valuable conclusion for him.¶
1.)-----------------------------preparation------------------------¶
1.a-- read the file and check the headers, status and import the library we need.¶
In [1]:
library(tidyverse)
library(pivottabler)
library(plotly)
library(psych)
install.packages("ggplot2::")
options(scipen=999)
In [2]:
mov <- read.csv("~/project-ionic/R/week4/movies.csv")
head(mov)
names(mov)
summary (mov)
str(mov)
2.)-------------------Four aspects----------------------------------------¶
2.a-1 -- Year, Genre and rating Distribution¶
-- First, check the number of movies per year¶
In [3]:
year_num <- mov %>% group_by(year) %>% summarize(num = n())
p<- plot_ly(year_num, x = ~year, y = ~num, name = 'Tree 1', type = 'scatter',
mode = 'lines')
p
- The number of movies after 1990 are the same in data.¶
In [4]:
with(mov, plot_ly(iris, x = genre, y= rating, z = year,
size = year, color = genre,
type="scatter3d", mode="markers"))
In [5]:
my_cols <- c("purple", "violetred1", "green3","red", "cyan", "steelblue","white","orange","pink","gray","black","maroon",
"coral","forestgreen","darkblue","lavender","ivory")
pairs(mov[,1:5], pch = 19, cex = 0.5,
col = my_cols[mov$genre],
lower.panel=NULL)
In [6]:
library(RColorBrewer)
barplot(table(mov$runtime,mov$score),col = brewer.pal(4,"Set1"))
In [7]:
genrePie <- mov %>% group_by(genre) %>% summarize (num = n())
genrePie
data=c(1331,392,277,359,2080,522,1444,14,32,277,4,38,15,13,18,2,2)
pct = (data/sum(data))*100
pct = round(pct,2)
labels = c("Action", "Adventure", "Animation",
"Biography", "Comedy","Crime","Drama","Family ","Fantasy","Horror","Musical ","Mystery","Romance"
,"Sci-Fi","Thriller","War","Western")
labels = paste(labels,pct, "%")
col = c("purple", "violetred1", "green3","red", "cyan", "steelblue","white","orange","pink","gray","black","maroon",
"coral","forestgreen","darkblue","lavender","ivory")
pie(pct,col = col, radius = 0.8, init.angle = 180, clockwise = TRUE,
labels =labels, main = "Genre Percentage")
In [8]:
plot_ly(genrePie, x = ~genre, y = ~num, type = 'bar', name = 'Sepal.Width') %>%
layout(yaxis = list(title = 'Count'), barmode = 'stack')
-- The Distribution above shows the most movies in our data belongs to the rating group : PG-13 and PG and the genre group: Action, Animation and comedy. This is a pie chart below to expain detaily So, we can say the Drama, Crime, Comedy, and Action is the mainstream of movies in the last 30 years.¶
2.b --Relationship of Score and Budget¶
In [9]:
head(mov[order(mov$score, decreasing = T), ], 10)
mean(mov$score)
plot( mov$score,mov$budget, type = "point",
pch=21, bg=c("yellow"), xlab = "Score", ylab ="Budget",
main="score and budget")
-- In here, there is some high budget movies with good score but we understand the popular movies not need high buget. By the way, the average of score is 6.5.¶
2.c --Relationship of Score and Votes¶
In [10]:
head(mov[order(mov$vote, decreasing = T), ], 10)
mean(mov$votes)
qplot(score, votes, data = mov, geom = c("point", "line"), color = I("red"))
- According to this chart whose line move forward gradually, we can know the vote from people will decide your score of movie really.¶
2.d --Relationship of Score and Gross¶
In [11]:
head(mov[order(mov$gross, decreasing = T), ], 10)
mean(mov$gross)
qplot(score, gross, data = mov, geom = c("line"), color = I("blue"))
- We learn from here, gross also like vote will affect the score. According the data above, we understand two factors which are uncontrollable will affect your socore. So, next we will start to predict the popular type of movie by this two factor. On the other hand, Budget is also a important thing of profit. We will anaylze later.¶
3.) --------------------Steven Spielberg's movies------------------------¶
-This will let us understand our director. Make a subset of Steven Spielberg¶
In [12]:
Ste <- mov %>% filter(mov$director == "Steven Spielberg")
head(Ste[order(Ste$score, decreasing = T), ], 10)
pairs.panels(Ste[,-5],
method = "pearson", # correlation method
hist.col = "#00AFBB",
density = TRUE, # show density plots
ellipses = TRUE # show correlation ellipses
)
3.a -- The total number of genre and score.¶
In [13]:
qhpvt(Ste, "score", "genre", "n()")
qplot(genre,score, data = Ste,
geom = c("point", "line"))
## 3.b -- The total number of company and genre.¶
In [14]:
qhpvt(Ste, "company", "genre", "n()")
DreamW <- mov %>% filter (company == "DreamWorks" )
mean(DreamW$score)
ggplot(DreamW, aes(x=year,y=score,color=genre)) + geom_point(alpha=0.5) + coord_fixed() + labs(title="DreamWorks",
x ="Year", y = "Score")
plot_ly(data = DreamW, x = ~year, y = ~score, color = ~genre)
universalP <- mov %>% filter (company == "Universal Pictures" )
ggplot(universalP, aes(x=year,y=score,color=genre)) + geom_point(alpha=0.5) + coord_fixed()+ labs(title="Universal Pictures",
x ="Year", y = "Score")
plot_ly(data = universalP, x = ~year, y = ~score, color = ~genre)
- we also check the score of historic movies in the big companies DreamWorks and Universal Picture. the average of two company almost same.¶
3.c -- The total number of genre and country.¶
In [15]:
qhpvt(Ste, "country", "genre", "n()")
3.d -- The average of runtime.¶
In [16]:
qhpvt(mov, "runtime", "score", "n()")
ggplot (Ste, aes (x = score, y = runtime, colour = score)) + stat_density2d ()
- In this section, we understand Mr. Spielberg have done 21 movies (20 in USA, 1 in France) in last 30 years. He has nine movies which are over 7.5 score. And he likes to work with DreamWorks and Universal Pictures. In addition, the average of runtime of his high score movies is around 150 mins. Two big companies cooperated with Mr. Spielberg are good choise to cooperate again.¶
4.) ------------------- Segmentation---------------------------------¶
- We will use the result we got from above to analyze into segmentations.¶
4.a - First, we anayze the high score movies in last 30 year¶
In [17]:
head(mov[order(mov$score, decreasing = T),], 10)
4.b - There is a lot of movies are too old. on the top, the time of movie is 1994, so I decide to analyze the data after 2010.¶
In [18]:
YMov <- filter(mov, year >= "2010" )
head(YMov[order(YMov$score, decreasing = T),], 10)
4.c - According the rank of score, the movies on the rank are from Warner Bros, Aamir Khan Productions, Quad Productions, and Amuse comanies. Because the Mr. Spielberg will produce a new movie. I recommand to anayze the data in two conunties which are have more movies data in here: USA and UK¶
In [19]:
YCMov <- filter(mov, year >= "2010" & (country =="USA" | country =="UK" ) )
summary(YCMov)
str(YCMov)
head(YCMov[order(YCMov$score, decreasing = T),], 10)
4d. - we find the data which is after 2010 and which in USA or UK. then we know the average of score is 6.4, but, we want to find the better data, we will filter out the movies whose score are lower than 8.¶
In [20]:
YCSMov <- filter(mov, year >= "2010" & (country =="USA" | country =="UK" ) & score >= "7")
head(YCSMov[order(YCSMov$score, decreasing = T),], 10)
4e. - Then, base on this data, we want to find the gross is over, Budget because the if the gross is not enough to pay the budget, it means there is no profit from movie that is not a good movies.¶
In [21]:
YCSGMov <- filter(mov, year >= "2010" & (country =="USA" | country =="UK" ) & score >= "8" & gross >= budget)
head(YCSGMov[order(YCSGMov$score, decreasing = T),], 10)
qhpvt(YCSGMov, "company", "score", "n()")
5.) ----------Best choose for Steven Spielberg-------------¶
5a. -- Best Movie Production Company¶
--After we filter out the movies whose score are under 8, we can find which company are good recently.¶
In [22]:
install.packages("ggrepel")
library("ggrepel")
In [23]:
YCSGCMov <- YCSGMov %>%group_by(company) %>% summarize ("total" = n () )
ggplot(YCSGCMov) + geom_bar(stat = "identity", color = 'steelblue', aes(x = company, y = total)) + theme(axis.text.x=element_text(size=rel(1), angle=90))
- Twentieth Century Fox Corporation have four over 8 score movies during 2010 ~ 2016. Wamer Bros also have three.¶
5b. - Best Writer¶
- A good movie are along with a good writer. We find Craig Borten and Jonathan Nolan are the good writers in USA and UK¶
In [24]:
YCSGWMov <- YCSGMov %>%group_by(writer) %>% summarize ("total" = n () )
ggplot(YCSGWMov, aes(x = writer, y = total, group = 1)) + geom_point(colour="steelblue") + geom_line() +theme(axis.text.x=element_text(size=rel(1), angle=90))
5c. Best Star¶
- A popular star will be contributed to a movie. So, let's analyze the star who has more popular movies.¶
In [25]:
YCSGSMov <- YCSGMov %>%group_by(star) %>% summarize ("total" = n () )
qhpvt(YCSGMov, "star", "score", "n()")
ggplot(YCSGSMov, aes(x = total, y = star, group = 1)) + geom_step(colour="orange") + geom_count(colour="pink") +theme(axis.text.x=element_text(size=rel(1), angle=90))
In [26]:
select_Leona = subset(mov, star == "Leonardo DiCaprio", select = c(name, score, star, year))
head(select_Leona[order(select_Leona$score, decreasing = T), ], 10)
qhpvt(select_Leona, "year", "score", "n()")
In [27]:
select_Matthew = subset(mov, star == "Matthew McConaughey", select = c(name, score, star, year))
head(select_Matthew[order(select_Matthew$score, decreasing = T), ], 10)
qhpvt(select_Matthew, "year", "score", "n()")
In [28]:
plot_ly(alpha = 0.6) %>%
add_histogram(x = ~select_Matthew$score, name ="Matthew") %>%
add_histogram(x = ~select_Leona$score, name = "Leona") %>%
layout(barmode = "overlay")